Beyond cardiac risk factors: non-cardiovascular comorbidities in sudden cardiac death prediction

TL;DR

This paper reviews how non-cardiovascular conditions like epilepsy and COPD significantly increase the risk of sudden cardiac death and should be included in prediction models.

Contribution

The paper highlights the underutilization of non-cardiovascular comorbidities in SCD prediction and proposes solutions for integrating them into models.

Findings

Neurologic and respiratory conditions increase SCD risk through autonomic and inflammatory mechanisms.

Non-cardiac comorbidities predict SCD and initial cardiac rhythm, affecting treatment decisions.

Current models lack non-cardiac conditions due to data and methodological challenges.

Abstract

Sudden cardiac death (SCD) causes 180,000–360,000 annual deaths in the United States, with mortality rates exceeding 90%. Despite advances in resuscitation science, predicting SCD remains challenging due to inconsistent definitions, subtle warning signs, and temporal variability in risk factors. While traditional cardiovascular conditions are well-integrated into risk prediction models, non-cardiovascular comorbidities remain significantly underutilized despite contributing to nearly 40% of SCD cases. This review examines evidence linking various systemic conditions to SCD risk. Neurologic disorders including epilepsy (1.6–5.89-fold increased risk), depression (1.6–2.7-fold), and anxiety (1.6-fold) elevate SCD vulnerability through autonomic dysregulation and medication effects. Respiratory conditions like COPD (1.3–3.6-fold) and obstructive sleep apnea (1.6–2.6-fold) contribute through chronic hypoxemia and inflammation. Hepatic pathology, kidney disease, anemia, and endocrine disorders (particularly diabetes with 1.7–2.4-fold risk) also demonstrate significant associations. Critically, non-cardiovascular comorbidities predict not only SCD occurrence but also initial cardiac rhythm presentation—essential for determining implantable cardioverter-defibrillator candidates, as these devices only benefit shockable rhythms. Conditions like epilepsy, depression, COPD, liver cirrhosis, and chronic kidney disease correlate with predominantly non-shockable presentations. Current prediction models incorporate few non-cardiac conditions, primarily due to historical cardiac-centric approaches, sample size constraints, complex disease interactions, and overfitting concerns. Proposed solutions include multidisciplinary research collaboration, multicenter data pooling, and advanced machine learning techniques to develop more comprehensive and accurate SCD prediction algorithms.

Figures